When uploading data for a survey, you can select a Quality Assurance (QA) setting that determines the quality threshold for processing.
There are four levels: Strict, Moderate, Lenient, and Skip all QA. These settings define the acceptable error for Ground Control Points (GCPs) and how the current survey is compared to previous ones. A stricter threshold means a lower tolerance for error, which could lead to manual review by a specialist if the data exceeds the set limits. The Skip all QA option provides the fastest processing time by automating the workflow and bypassing manual checks, leaving the quality assessment to you. Your choice allows you to balance the need for accuracy with processing speed, ensuring the final output meets your specific requirements.
What is Ground Control Point (GCP) error?
Data processing
After you submit your survey, it enters our processing pipeline, where a complex feature-matching process occurs. Here, an algorithm detects similar points in overlapping photos and creates a sparse point cloud. A computer will georeference the sparse point cloud by marking ground control points (GCPs) in the source imagery, which adjusts the model. A geospatial specialist will intervene if the computer doesn’t mark any points, and those points will be manually marked.
After the point cloud has been georeferenced, a root mean square error (RMSE) calculation is performed to get the distance between the GCP's original input position and the estimated position of the same GCP after the adjustment is made. This distance is the GCP error and is displayed within the Expected Ground Control Accuracy table in the Processing Report.
Strict, Moderate, Lenient, or Skip all QA
The user’s choice—Strict, Moderate, Lenient, or Skip all QA—will determine what RMSE value a survey will be flagged and placed on hold. The higher the allowable GCP error, the more likely the survey will flow through our processing pipeline without delays.
The RMSE thresholds are:
| Option | RMSE Threshold |
| Strict | 5 cm (0.16 ft) |
| Moderate | 10 cm (0.33 ft) |
| Lenient | 20 cm (0.66 ft) |
| Skip all QA | No data review. Users can review GCP errors in the Processing Report. |
A geospatial specialist will perform additional photogrammetry troubleshooting if any GCP within the Expected Ground Control Accuracy table has an RMSE above the set threshold. If the error is not resolved, a geospatial specialist will place the survey on hold and contact the user who uploaded it with an explanation for:
- Why this is occurring, and
- Alternative options for processing
What about comparing to past surveys?
The final step of the Propeller processing pipeline is to compare the submitted survey with past overlapping surveys.
Strict
When the box is checked and Strict is selected, if more than 50% of the survey area has changed by more than 0.10 m (0.33 ft), the survey is flagged and requires the intervention of a geospatial specialist. The geospatial specialist will release the survey if the changes are due to legitimate site activity. If the changes are not due to site activity and are observed in areas that appear not to have changed, then the geospatial specialist will place the survey on hold and reach out to the user who uploaded the data.
Moderate
Similarly, when the box is checked, and Moderate is selected, a similar process occurs, but it will only occur if more than 60% of the survey area has changed by more than 0.20m (0.66 ft).
Lenient
When the box is checked and Lenient is selected, the survey will be flagged if more than 60% of the survey area has changed by 0.40m (1.31 ft).
Skip all QA
The survey will not be compared to the previous survey. The user can make comparisons using a Volume Compare measurement.
Please note that this compare-to-past-surveys process is not mandatory and can be unchecked by the user. This means we will not compare the current survey with previous ones, which will result in a quicker turnaround time.
Selecting a Comparison Survey
The photogrammetry uploader now provides direct control over which previous survey is used for comparison during processing, allowing for precise evaluation against past terrain data.
This capability prevents unintended comparisons that could lead to processing holds, contributing to more accurate results and faster turnaround times by reducing manual troubleshooting by Geospatial Specialists.
While ideal comparison surveys share medium-to-high overlap, consistent QA, historical alignment, and matching processing methods, the optimal selection may vary depending on a user's specific use case. Users will be guided by the information displayed in the dropdown to make the best choice for their needs.
Understanding Comparison Survey Details
When selecting a comparison survey from the dropdown menu, you'll find several columns providing key information to help you choose the most appropriate historical data.
Each column indicates the following:
- Survey name: The name of the historical survey available for comparison.
-
Approx. survey overlap: This value describes the approximate percentage of overlap between the boundary of your current survey (derived from camera locations) and the boundary of the historical survey.
Surveys with images taken directly overhead (nadir shots) generally provide a more accurate overlap approximation.Surveys with images taken at an angle (oblique shots) may result in a less accurate approximation.
Keep in mind that this is an approximation to guide your selection, and inaccuracies can occur, especially for surveys with fewer flight lines like corridor surveys.
- High overlap: The camera locations from the current upload have a significant approximate overlap with those in this survey.
- Medium overlap: The camera locations from the current upload have a partial approximate overlap with those in this survey.
- Low overlap: The camera locations from the current upload have a minimal approximate overlap with those in this survey.
- QA level: The Quality Assurance (QA) level (Strict, Moderate, Lenient, or Skip all QA) that was applied during the processing of the historical survey.
- Capture date: The date the historical survey was captured.
- Compared to past survey: This column indicates whether the historical survey itself was also compared to an even older survey and was deemed to be historically aligned during its processing.
If the option to compare to past surveys is unchecked, the survey will process without a comparison.
Slider Behavior
The Slider will default to Lenient if the data provided in the uploader fails any initial quality checks. These checks include the following:
For these initial quality checks, if Lenient is recommended and the user opts to change the slider to Moderate or Strict, it is likely the survey will be placed on hold for investigation by our geospatial specialists, resulting in a longer turnaround time.
Lenient QA required
If the data fails quality checks and warns you that you are not flying a compatible drone and camera, this will force the slider to Lenient and not allow you to change it.
Skip all QA required
If the data fails quality checks and warns you of the use of standard accuracy and with no GCPs, this will force the slider to Skip all QA and not allow you to change it.
More about skipping all QA
This option has the fastest turnaround time as it will send the survey through Propeller’s automated processing workflow. A computer will automatically mark aeroPoints and GCPs, and the survey will not be held for review by a GIS specialist for any reason.
Some GCPs may be excluded if they are not easily recognizable by a computer. We find the best results are achieved with AeroPoints or other checkerboard-style GCPs when placed on uniformly colored ground. However, the computer will still mark other styles of GCP markers. Please see our best practices for GCP markers in our article on How to Use Ground Control.
Use the Processing Report and Volume Compare measurements to assess the quality of the survey and ensure that it meets your organization’s survey requirements.
I still can't do it!
We wrote these articles to equip you with everything you need to get the job done on your own, but we understand that sometimes this isn't sufficient.
If you're stuck, you can connect with our support team by clicking the question mark button at the top right corner of your user portal.
Related to